1. Field of the Invention
This invention generally relates to sorting objects, and more particularly to a system and method which uses a voice-recognition unit to aid an operator in sorting letters, parcels, and other items of mail.
2. Description of the Related Art
Sorting systems have been used for decades by public and private organizations to route mail from one location to another. These systems generally take one of two approaches; they are either manual in nature or automated.
Manual sorting requires a worker to read name and address information on each piece of mail before placing it by hand into one of a number of bins. This approach has proven to be slow and inefficient and thus unsuitable for use in high-volume applications. Further, this kind of sorting tends to be inaccurate because it relies on the worker""s memory and concentration skills. In order to sort, for example, a worker must have a fore knowledge of the mail routes and the bins corresponding to each route. And, if the worker forgets this information or is unfamiliar with a name or address, he must reference sorting tables, which further slows the process.
Automated sorting systems tend to be faster and more efficient than manual sorting techniques. This improved performance is achieved mainly through the use of a workstation situated between a feeder and sorting machine. The workstation typically includes a minimum of a processor, display terminal, database of addressee information, and keyboard. In operation, the feeder sends a letter to the workstation whereupon an addressee name on the letter is input into the system. The processor searches the database to locate a record matching the name, and the sorting machine then sorts the letter into a bin based on information in the matching record.
Improvements in automated sorting systems have principally focused on the input and output of information at the workstation. In early systems, information input was performed by an operator-controlled keyboard. See, for example, U.S. Pat. No. 4,578,759. Because keyboard entry tends to be slow, systems of this type are undesirable, especially for high-volume applications. Other systems use optical scanners to read the information on mail. See, for example, U.S. Pat. Nos. 4,358,017, 4,503,977, and 4,641,753 each of which discloses a system that reads computer-generated marks, e.g., bar codes. Another system, disclosed in U.S. Pat. No. 4,632,252, uses optical character recognition techniques to input names and addresses. While these systems sort letters at a potentially faster rate than keyboard-based systems, their use is limited to only sorting mail having computer-generated marks.
Still other systems are multiple-input systems, i.e., they accept input information from more than one device. U.S. Pat. Nos. 3,587,856, 4,307,293, and 4,921,107 disclose systems of this type which accept either keyboard or optical scanner input depending on the operator""s choice or upon whether the letters have computer-generated codes.
Another type of multiple-input system allows an operator to input addressee information via a keyboard and voice-recognition unit. In this system, disclosed in U.S. Pat. No. 4,921,107 to Hofer, when letters are transported to a viewing position of a workstation, an operator speaks the addressee""s name into a microphone. The voice-recognition unit translates the speech into digital signals identical to a keyboard input, and the workstation processor then controls an automatic sorting machine to sort the letter based on destination information in a matching database record.
Voice-driven systems like Hofer represent an improvement in the art because they allow addressee information to be input in a hands-free manner. Nevertheless, these systems have significant drawbacks. For example, the Hofer system sorts letters using a conventional sorting machine, for example, as described in U.S. Pat. No. 4,307,293 to Lazzarotti. Sorting machines of this type have a finite number of bins of limited capacity. Consequently, the volume they are able to handle is limited. Also, the voice-recognition unit of Hofer, made by Kurzweil Artificial Intelligence Corporation, is a primitive, hardware-based speech-recognition circuit which is costly by today""s standards and often inaccurate.
In addition to the individual shortcomings discussed above, conventional automated sorting systems all have at least one additional drawback: They require an operator to be confined to a workstation. This is true even in optical scanning systems since the operator if required to make decisions, for example, when conflicts occur that can only be resolved through keyboard entry.
U.S. Pat. No. 5,677,834 to Mooneyham discloses a system which attempts to overcome this drawback by combining the manual and automated sorting approaches discussed above. The Mooneyham system does this by including a body-supported computer linked to a headset which functions as both an input and output unit. As shown in FIGS. 11A and 11B, computer 70 is adapted for attachment to a user""s belt and includes a database for storing addressee information 71, a processor 72, a voice synthesis unit 73, a voice recognition unit 74, an input keypad 75, and a display 76. Headset 77 is equipped with a microphone 78 and speaker 79 and is connected to the portable computer by an electrical cord.
In operation, a user is presented with a number of mail items to be sorted. One by one, the user announces a street address into the headset microphone. The voice recognition unit recognizes the address and the processor searches the database for a matching record. If a match is found, the voice synthesis unit is activated to output destination information (e.g., a delivery route or bin number) to the user through the headset speakers.
The Mooneyham system has a number of drawbacks which make it undesirable from an efficiency standpoint. First, Mooneyham encumbers its user""s mobility by requiring him to wear essentially the entire workstation on his body. This makes the system susceptible to physical damage and excessive wear, as the user may be expected to be very active within a confined area and thus the computer may knock into objects or be dropped on the floor during sorting.
Second, the information database must be limited in size in order to fit within the body-worn computer. As a result, the number of records stored in the database is small in comparison to conventional workstations. Also, the Mooneyham computer is a completely stand-alone system and therefore cannot access multiple databases at once or other databases through a network connection, all of which further limits functionality.
Third, when an error occurs or the voice recognition unit is unable to understand a microphone input, the user is required look at the portable computer display on his belt. This diverts the user""s attention away from sorting, slowing the sorting process.
From the foregoing discussion, it is clear that a need exists for an improved, voice-driven system and method for assisting an operator in sorting items of mail without requiring the operator to be confined to a workstation, and more particularly a system and method which provides the operator with maximum mobility without sacrificing processing power, memory capacity, and system functionality.
It is a first object of the present invention to provide a system and method for assisting an operator in sorting items of mail that gives the operator virtually unlimited mobility, while at the same time maximizing system processing power, memory capacity, and functionality.
It is a second object of the present invention to achieve the first objective by providing a headset equipped with a microphone which allows the operator to wirelessly transmit addressee information to a workstation.
It is another object of the present invention to increase the range of the wireless headset so that it may be operated from virtually any distance, thereby allowing the workstation to be remotely located from the operator, if desired, and even on different rooms or floors.
It is another object of the present invention to provide a system and method of the aforementioned type which allows the operator to simultaneously access multiple databases at a time, and/or databases at other workstations or locations through a network connection.
It is another object of the present invention to incorporate a speaker into the headset which outputs destination information to the operator in a synthesized voice, thereby eliminating the need to read this information from a workstation display and giving the operator even greater mobility during sorting.
It is another object of the present invention to incorporate a heads-up display into the headset which displays destination information to the operator, as well as a series of interactive screens which allow the operator to re-configure or otherwise control the system through voice-drive commands, further freeing the user from the confines of the workstation.
It is another object of the present invention to provide a system and method which performs voice-recognition with improved accuracy compared with conventional mail sorting systems equipped with voice-recognition units.
It is another object of the present invention to increase the functionality of the workstation of the present invention over conventional workstations by providing a pronunciation engine, a statistical tracking feature, a report generation feature, a quick-search feature, as well as other functions which improve the management of sorting operations, all of which are either selectable or controllable by the headset of the present invention.
It is another object of the present invention to automatically print a label bearing destination, addressee, and/or other information when a database search has found a matching record, which label further streamlines the sorting process by giving personnel a guide as to the distribution of mail items.
These and other objectives of the invention are achieved by providing a system and method which uses a wireless headset adapted to communicate with a fully functioning workstation. In a first embodiment, the headset is equipped with a microphone which converts addressee information and commands into speech signals which are then transmitted to a voice recognition unit in the workstation. The voice recognition unit is preferably software-driven, speaker-independent, and supports multiple operators using adaptively trained models. Once recognized, addressee information is used by a processing unit to search a database of records. Destination information in a matching database record is then output to a display terminal where the operator is located and sorting is performed accordingly.
In a second embodiment, a speaker and receiver are included in the headset to sustain duplex communications, and a voice synthesis unit is connected to the processing unit of the workstation. When a database search is performed, the voice synthesis unit converts destination information in a matching record into speech signals that are transmitted to the headset. The speaker in the headset then audibly communicate the information to the user.
In a third embodiment, a heads-up display and receiver are included in the headset. When a database search is performed, the destination information in a matching record is transmitted to the heads-up display, which is then viewed by the operator for sorting.
The second and third embodiments are especially advantageous since the headset speakers and heads-up display each allow the user not only to sort mail items without interruption, but also to select between addressees when more than one matching record is found in the search. Also, through the heads-up display, the operator can access interactive screens for re-configuring the system or performing any number of optional system functions without having to return to the workstation site. This is especially advantageous when the workstation is remotely located from the mail sorting area.
Optional system functions include volume and operator productivity report generation, statistical analysis and graphing, speak or spell options, network connection capability, and database import capability. The system also includes a pronunciation unit which assists an operator in correctly pronouncing addressee information, and a printer is provided for printing labels containing destination or other information. If desired, the printer may be controlled to automatically print a label when database results are returned to the headset.